Threshold segmentation is a basic grey-scale converting technology of image processing. We combine threshold segmentation with Chan-Vese model with an accompanying scale transformation to improve the segment speed and effect. Experimental results show that this program greatly enhances the convergent speed of the Chan-Vese model，and it focuses on changing the defects of more iterative times and low diffusion speed of Chan-Vese model on the border of the image when its grayscale changes slowly.